Retinal vessel segmentation using supervised classification based on multi-scale vessel filtering and gabor wavelet

We present an automated segmentation method for blood vessels in images of the ocular fundus. The method uses a supervised classification of vessels at each pixel based on its feature vectors. The feature vectors include the responses of the pixel to the multi-scale vessel enhancement filtering and Gabor filtering at multiple scales and multiple orientations. We use a support vector machine to extract the vessels. The performance of the proposed method is evaluated on a DRIVE database. The accuracy of the vessel segmentation reaches more than 95%, which indicates the effectiveness of the proposed method.